Anthropic AI Updates: October 21, 2026

1. Claude Designs Protein Binders Against Scorpion Neurotoxins, 17 of 211 Confirmed in the Lab

Anthropic. Anthropic reported that an internal research Claude model, working mostly autonomously for 64 hours within Claude Science, designed minibinders against scorpion neurotoxins in a project developed with Citrini Research. Claude generated 318,484 protein designs and selected 180 for testing, the team added 31 more, and contract research organizations Adaptyv Bio and Twist Bioscience lab-tested all 211, with 17 meeting the binding criteria across four neurotoxins dangerous to humans. Anthropic states it has not confirmed whether the binders neutralize venom in a cellular environment and does not claim to have produced an antivenom. All 211 designs and their data are being released alongside a technical report. Source

2. Claude Code 2.1.305 Lets Sandboxed Linux Commands Create User Namespaces

Anthropic. Claude Code 2.1.305 adds a sandbox.allowNestedUserNamespaces setting so sandboxed commands on Linux can create user namespaces, which tools such as Chromium and rootless podman require; the changelog notes this lets those commands get around the sandbox’s file protections. A new CLAUDE_CODE_RETRY_WATCHDOG_USAGE_LIMIT_RECHECK_MS variable lets unattended retry mode notice when a usage limit is lifted before its scheduled reset. The release also adds a “Running start-up hooks” spinner status in VS Code and lets Claude Tag admins find public Slack channels Claude has not yet joined when adding a place. Source